FlytBase Blog
Cycle Counting Using Drones

Cycle Counting Inventory in Warehouses Using Autonomous Drone Fleets

Redefining Warehouse Inventory Accuracy Metrics & Workflows with Drone-based Cycle Counts

Inventory accuracy, in the context of cycle counting of warehouse inventory, tends to be a somewhat subjective term. This is because (in)accuracy could involve one or more of the following issues:

  • Was an item in its expected location, but the counting activity did not capture it?
  • Was the item missing from its expected location – either missing from the entire warehouse or placed in an incorrect location?
  • Was the item that was not found in the expected location relatively ‘valuable’?

Thus, inventory inaccuracy can be viewed from multiple perspectives: operations, profit & loss, capital-at-risk, annual audits, etc. Depending on the strategic imperatives of a given warehouse business, the inventory management stakeholders have to prioritize amongst:

  • Frequency/interval of cycle counts
  • Number of resources (people, equipment) deployed for cycle counts
  • Coverage of cycle counts
  • Accuracy/variance of cycle counts
  • Opportunity cost of cycle counts (revenue foregone due to shutdowns, etc.)

Real World vs. Digital Twin

Cycle counting can be thought of as a continuous effort to make the warehouse management system (i.e. the digital twin) reflect the item↔location mapping as it actually is in the warehouse (i.e. the real world).

cycle counting in warehouseThe frequency of cycle counts will determine how often the ‘delta’ (i.e. the mismatch or gap) between reality and records is resolved. Weekly cycle counts, instead of monthly, will minimize this gap, but will cost extra time, effort, resource expenditure – and potentially foregone revenue. The danger of not counting inventory frequently enough is the compounding of the error in inventory records stored in the warehouse management system.

 accuracy of cycle countsThe accuracy of cycle counts will determine how low – ideally zero! – can the ‘delta’ be pushed down to. Of course, the inherent nature of cycle counting implies that a zero gap is impossible – but the more accurate the count, the lesser the gap – and thus a positive impact on metrics related to buffer stock, order fulfillment, stockouts, etc.
improving accuracy of cycle countsIn fact, it could be argued that even if drone-based inventory counts result in a small loss of accuracy versus manual counts, the ability to do more frequent, safer, autonomous counts with drones will result in greater (net) business value!

Coverage vs. Accuracy

Ideally, warehouse executives would want 100% coverage of inventory during counts i.e. every single location must be scanned, even if it takes more time, extra resources and may result in inaccurate scans due to human error. The goal here is 100% coverage.

100% accuracy, on the other hand, implies checking whether – of whatever was covered (100% or otherwise) – were all the items mapped to their correct locations? For instance, if high, medium and low-value items are stored separately, managers may choose to forego 100% coverage of all items and instead want to ensure 100% accuracy of (at least) the high-value items.

Thus, trading off coverage for accuracy is a managerial decision, in the context of revenue targets, cost pressures, audit procedures and operational constraints. Cycle count methods & best practices have evolved to help managers analyze these tradeoffs and optimize physical inventory accuracy.

Drone-based Cycle Count Automation

Viewed simplistically, drones used for cycle counting can do more or less what humans (with handheld barcode scanners and ladders/trucks/lifts) can do. Autonomous drone fleets can, no doubt, enable faster, cheaper and safer indoor inventory counts – but what about accuracy & other metrics?

The journey of drone adoption in effective inventory management tends to begin with proof-of-concept (PoC) projects, that mature into pilot programs and eventually into production deployments across all sites. While considering a PoC for drone-based inventory scans, a common misconception that tends to crop up is that drones can completely replace manual stock takes. This is neither feasible nor intended – drones are just another tool in the warehouse automation toolkit – albeit a powerful one for minimizing inventory variance and cycle counting issues.

Manual vs. Drone-based Cycle Counts

The table below lists the major sources of inventory inaccuracy – the common factor being the lack of traceability of the source of inaccuracy in manual cycle counts.

Sources of Inventory Inaccuracy Manual Cycle Count Drone Cycle Count
 1 Hard-to-reach locations inaccessible by a person  ☹️  😃
 2 Scanning of an incorrect (item or location) barcode by a person  ☹️  😃
 3 Intentional skipping of an item or location by a person  ☹️  😃
 4 Theft by a person during stock takes  ☹️  😃
 5 Errors in manual entry of cycle count data into WMS  ☹️  😃
 6 Unreadable (damaged, plastic-covered, etc.) or missing barcode  ☹️  😐︎

In manual cycle counts, it is impossible to determine the exact source of a given error in the item↔location mapping, since there is no traceability of the manual scan. The only way to check is to repeat the entire process!

Drones – on the other hand – will either:

  • Correctly read the item & location barcodes, or
  • Provide location-wise image & video data, thus enabling a human to physically go to the location to analyze/fix the error.

Thus drones provide full traceability – serving as pointers to locations of inventory inaccuracy, where human intervention can then be used in a planned manner. How to control inventory variance, and resolve inventory discrepancies, is the next challenge – but one that is made easier with drone adoption.

Drone-first, Manual-later Cycle Counts

Thus, an ‘and’ approach is the wiser one, as opposed to an ‘or’ approach when considering drones in warehouses. A drone-first cycle counting workflow can be designed with the following steps:

  1. Select warehouse zones that are most suitable for drones (rack-pallet storage, no human activity in aisles, 1-deep pallets with front-facing barcodes).
  2. Determine the cycle counting frequency, and time available each day/week/month for drone flights.
  3. Determine the optimal balance between speed of cycle counting and capital expenditure (on drone kits and charging infrastructure).
  4. Train an operator to oversee drone operations, replace batteries (if autonomous charging isn’t suitable), and monitor data imports into WMS.
  5. Plan & execute autonomous drone missions, starting with a PoC deployment in an aisle in one warehouse.
  6. At the end of each drone-based cycle count, review drone data, and:
    • View images, video of each location where the drone solution raised an alert, and manually update the item/location barcode,
    • Physically inspect locations where the image, video were insufficient to help resolve the alert,
    • Involve warehouse stakeholders to analyze and address the root-cause,
    • Update the drone data capture template, workflow, etc. as necessary.

Summary

Cycle counting of inventory is a key lever in operational and capital efficiency of warehouse operations. Autonomous drone fleets offer a cost-effective, reliable, scalable solution for creating business value – specifically for cycle counts. Given the relatively low capital investment, drones are ideal for starting the warehouse automation journey. However, such drone operations involve tradeoffs between humans-in-the-loop and humans-outside-the-loop of stock takes. The most effective warehouse drone operations will adopt physical counting workflows that use a drone-first approach, and augment inventory management operations using human resources to target 100% coverage and 100% inventory accuracy.

With the launch of the FlytWare Proof-of-Concept package, you can immediately adopt drones for inventory cycle counting.

Write to us at info@flytbase.com or schedule a call with the FlytWare team.


References

Add comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.